Home | History | Annotate | Line # | Download | only in INSTALL
specific.html revision 1.12
      1 <!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
      2 <html>
      3 <!-- Copyright (C) 1988-2019 Free Software Foundation, Inc.
      4 
      5 Permission is granted to copy, distribute and/or modify this document
      6 under the terms of the GNU Free Documentation License, Version 1.3 or
      7 any later version published by the Free Software Foundation; with no
      8 Invariant Sections, the Front-Cover texts being (a) (see below), and
      9 with the Back-Cover Texts being (b) (see below).  A copy of the
     10 license is included in the section entitled "GNU
     11 Free Documentation License".
     12 
     13 (a) The FSF's Front-Cover Text is:
     14 
     15 A GNU Manual
     16 
     17 (b) The FSF's Back-Cover Text is:
     18 
     19 You have freedom to copy and modify this GNU Manual, like GNU
     20      software.  Copies published by the Free Software Foundation raise
     21      funds for GNU development. -->
     22 <!-- Created by GNU Texinfo 6.5, http://www.gnu.org/software/texinfo/ -->
     23 <head>
     24 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
     25 <title>Installing GCC</title>
     26 
     27 <meta name="description" content="Installing GCC">
     28 <meta name="keywords" content="Installing GCC">
     29 <meta name="resource-type" content="document">
     30 <meta name="distribution" content="global">
     31 <meta name="Generator" content="makeinfo">
     32 <style type="text/css">
     33 <!--
     34 a.summary-letter {text-decoration: none}
     35 blockquote.indentedblock {margin-right: 0em}
     36 blockquote.smallindentedblock {margin-right: 0em; font-size: smaller}
     37 blockquote.smallquotation {font-size: smaller}
     38 div.display {margin-left: 3.2em}
     39 div.example {margin-left: 3.2em}
     40 div.lisp {margin-left: 3.2em}
     41 div.smalldisplay {margin-left: 3.2em}
     42 div.smallexample {margin-left: 3.2em}
     43 div.smalllisp {margin-left: 3.2em}
     44 kbd {font-style: oblique}
     45 pre.display {font-family: inherit}
     46 pre.format {font-family: inherit}
     47 pre.menu-comment {font-family: serif}
     48 pre.menu-preformatted {font-family: serif}
     49 pre.smalldisplay {font-family: inherit; font-size: smaller}
     50 pre.smallexample {font-size: smaller}
     51 pre.smallformat {font-family: inherit; font-size: smaller}
     52 pre.smalllisp {font-size: smaller}
     53 span.nolinebreak {white-space: nowrap}
     54 span.roman {font-family: initial; font-weight: normal}
     55 span.sansserif {font-family: sans-serif; font-weight: normal}
     56 ul.no-bullet {list-style: none}
     57 -->
     58 </style>
     59 
     60 
     61 </head>
     62 
     63 <body lang="en">
     64 <h1 class="settitle" align="center">Installing GCC</h1>
     65 
     66 
     67 
     68 
     69 
     70 
     71 
     72 
     73 
     74 
     75 
     76 
     77 
     78 
     79 
     80 
     81 
     82 
     83 
     84 
     85 
     86 
     87 
     88 
     89 <a name="index-Specific"></a>
     90 <a name="index-Specific-installation-notes"></a>
     91 <a name="index-Target-specific-installation"></a>
     92 <a name="index-Host-specific-installation"></a>
     93 <a name="index-Target-specific-installation-notes"></a>
     94 
     95 <p>Please read this document carefully <em>before</em> installing the
     96 GNU Compiler Collection on your machine.
     97 </p>
     98 <p>Note that this list of install notes is <em>not</em> a list of supported
     99 hosts or targets.  Not all supported hosts and targets are listed
    100 here, only the ones that require host-specific or target-specific
    101 information have to. 
    102 </p>
    103 <ul>
    104 <li> <a href="#aarch64-x-x">aarch64*-*-*</a>
    105 </li><li> <a href="#alpha-x-x">alpha*-*-*</a>
    106 </li><li> <a href="#amd64-x-solaris210">amd64-*-solaris2.10</a>
    107 </li><li> <a href="#arm-x-eabi">arm-*-eabi</a>
    108 </li><li> <a href="#avr">avr</a>
    109 </li><li> <a href="#bfin">Blackfin</a>
    110 </li><li> <a href="#dos">DOS</a>
    111 </li><li> <a href="#x-x-freebsd">*-*-freebsd*</a>
    112 </li><li> <a href="#h8300-hms">h8300-hms</a>
    113 </li><li> <a href="#hppa-hp-hpux">hppa*-hp-hpux*</a>
    114 </li><li> <a href="#hppa-hp-hpux10">hppa*-hp-hpux10</a>
    115 </li><li> <a href="#hppa-hp-hpux11">hppa*-hp-hpux11</a>
    116 </li><li> <a href="#x-x-linux-gnu">*-*-linux-gnu</a>
    117 </li><li> <a href="#ix86-x-linux">i?86-*-linux*</a>
    118 </li><li> <a href="#ix86-x-solaris210">i?86-*-solaris2.10</a>
    119 </li><li> <a href="#ia64-x-linux">ia64-*-linux</a>
    120 </li><li> <a href="#ia64-x-hpux">ia64-*-hpux*</a>
    121 </li><li> <a href="#x-ibm-aix">*-ibm-aix*</a>
    122 </li><li> <a href="#iq2000-x-elf">iq2000-*-elf</a>
    123 </li><li> <a href="#lm32-x-elf">lm32-*-elf</a>
    124 </li><li> <a href="#lm32-x-uclinux">lm32-*-uclinux</a>
    125 </li><li> <a href="#m32c-x-elf">m32c-*-elf</a>
    126 </li><li> <a href="#m32r-x-elf">m32r-*-elf</a>
    127 </li><li> <a href="#m68k-x-x">m68k-*-*</a>
    128 </li><li> <a href="#m68k-uclinux">m68k-uclinux</a>
    129 </li><li> <a href="#microblaze-x-elf">microblaze-*-elf</a>
    130 </li><li> <a href="#mips-x-x">mips-*-*</a>
    131 </li><li> <a href="#nds32le-x-elf">nds32le-*-elf</a>
    132 </li><li> <a href="#nds32be-x-elf">nds32be-*-elf</a>
    133 </li><li> <a href="#nvptx-x-none">nvptx-*-none</a>
    134 </li><li> <a href="#or1k-x-elf">or1k-*-elf</a>
    135 </li><li> <a href="#or1k-x-linux">or1k-*-linux</a>
    136 </li><li> <a href="#powerpc-x-x">powerpc*-*-*</a>
    137 </li><li> <a href="#powerpc-x-darwin">powerpc-*-darwin*</a>
    138 </li><li> <a href="#powerpc-x-elf">powerpc-*-elf</a>
    139 </li><li> <a href="#powerpc-x-linux-gnu">powerpc*-*-linux-gnu*</a>
    140 </li><li> <a href="#powerpc-x-netbsd">powerpc-*-netbsd*</a>
    141 </li><li> <a href="#powerpc-x-eabisim">powerpc-*-eabisim</a>
    142 </li><li> <a href="#powerpc-x-eabi">powerpc-*-eabi</a>
    143 </li><li> <a href="#powerpcle-x-elf">powerpcle-*-elf</a>
    144 </li><li> <a href="#powerpcle-x-eabisim">powerpcle-*-eabisim</a>
    145 </li><li> <a href="#powerpcle-x-eabi">powerpcle-*-eabi</a>
    146 </li><li> <a href="#riscv32-x-elf">riscv32-*-elf</a>
    147 </li><li> <a href="#riscv32-x-linux">riscv32-*-linux</a>
    148 </li><li> <a href="#riscv64-x-elf">riscv64-*-elf</a>
    149 </li><li> <a href="#riscv64-x-linux">riscv64-*-linux</a>
    150 </li><li> <a href="#s390-x-linux">s390-*-linux*</a>
    151 </li><li> <a href="#s390x-x-linux">s390x-*-linux*</a>
    152 </li><li> <a href="#s390x-ibm-tpf">s390x-ibm-tpf*</a>
    153 </li><li> <a href="#x-x-solaris2">*-*-solaris2*</a>
    154 </li><li> <a href="#sparc-x-x">sparc*-*-*</a>
    155 </li><li> <a href="#sparc-sun-solaris2">sparc-sun-solaris2*</a>
    156 </li><li> <a href="#sparc-sun-solaris210">sparc-sun-solaris2.10</a>
    157 </li><li> <a href="#sparc-x-linux">sparc-*-linux*</a>
    158 </li><li> <a href="#sparc64-x-solaris2">sparc64-*-solaris2*</a>
    159 </li><li> <a href="#sparcv9-x-solaris2">sparcv9-*-solaris2*</a>
    160 </li><li> <a href="#c6x-x-x">c6x-*-*</a>
    161 </li><li> <a href="#tilegx-x-linux">tilegx-*-linux*</a>
    162 </li><li> <a href="#tilegxbe-x-linux">tilegxbe-*-linux*</a>
    163 </li><li> <a href="#tilepro-x-linux">tilepro-*-linux*</a>
    164 </li><li> <a href="#visium-x-elf">visium-*-elf</a>
    165 </li><li> <a href="#x-x-vxworks">*-*-vxworks*</a>
    166 </li><li> <a href="#x86-64-x-x">x86_64-*-*, amd64-*-*</a>
    167 </li><li> <a href="#x86-64-x-solaris210">x86_64-*-solaris2.1[0-9]*</a>
    168 </li><li> <a href="#xtensa-x-elf">xtensa*-*-elf</a>
    169 </li><li> <a href="#xtensa-x-linux">xtensa*-*-linux*</a>
    170 </li><li> <a href="#windows">Microsoft Windows</a>
    171 </li><li> <a href="#x-x-cygwin">*-*-cygwin</a>
    172 </li><li> <a href="#x-x-mingw32">*-*-mingw32</a>
    173 </li><li> <a href="#os2">OS/2</a>
    174 </li><li> <a href="#older">Older systems</a>
    175 </li></ul>
    176 
    177 <ul>
    178 <li> <a href="#elf">all ELF targets</a> (SVR4, Solaris 2, etc.)
    179 </li></ul>
    180 
    181 
    182 <!-- -------- host/target specific issues start here ---------------- -->
    183 <hr />
    184 <a name="aarch64-x-x"></a><a name="aarch64*-*-*"></a>
    185 <h3 class="heading">aarch64*-*-*</h3>
    186 <p>Binutils pre 2.24 does not have support for selecting <samp>-mabi</samp> and
    187 does not support ILP32.  If it is used to build GCC 4.9 or later, GCC will
    188 not support option <samp>-mabi=ilp32</samp>.
    189 </p>
    190 <p>To enable a workaround for the Cortex-A53 erratum number 835769 by default
    191 (for all CPUs regardless of -mcpu option given) at configure time use the
    192 <samp>--enable-fix-cortex-a53-835769</samp> option.  This will enable the fix by
    193 default and can be explicitly disabled during compilation by passing the
    194 <samp>-mno-fix-cortex-a53-835769</samp> option.  Conversely,
    195 <samp>--disable-fix-cortex-a53-835769</samp> will disable the workaround by
    196 default.  The workaround is disabled by default if neither of
    197 <samp>--enable-fix-cortex-a53-835769</samp> or
    198 <samp>--disable-fix-cortex-a53-835769</samp> is given at configure time.
    199 </p>
    200 <p>To enable a workaround for the Cortex-A53 erratum number 843419 by default
    201 (for all CPUs regardless of -mcpu option given) at configure time use the
    202 <samp>--enable-fix-cortex-a53-843419</samp> option.  This workaround is applied at
    203 link time.  Enabling the workaround will cause GCC to pass the relevant option
    204 to the linker.  It can be explicitly disabled during compilation by passing the
    205 <samp>-mno-fix-cortex-a53-843419</samp> option.  Conversely,
    206 <samp>--disable-fix-cortex-a53-843419</samp> will disable the workaround by default.
    207 The workaround is disabled by default if neither of
    208 <samp>--enable-fix-cortex-a53-843419</samp> or
    209 <samp>--disable-fix-cortex-a53-843419</samp> is given at configure time.
    210 </p>
    211 <p>To enable Branch Target Identification Mechanism and Return Address Signing by
    212 default at configure time use the <samp>--enable-standard-branch-protection</samp>
    213 option.  This is equivalent to having <samp>-mbranch-protection=standard</samp>
    214 during compilation.  This can be explicitly disabled during compilation by
    215 passing the <samp>-mbranch-protection=none</samp> option which turns off all
    216 types of branch protections.  Conversely,
    217 <samp>--disable-standard-branch-protection</samp> will disable both the
    218 protections by default.  This mechanism is turned off by default if neither
    219 of the options are given at configure time.
    220 </p>
    221 <hr />
    222 <a name="alpha-x-x"></a><a name="alpha*-*-*"></a>
    223 <h3 class="heading">alpha*-*-*</h3>
    224 <p>This section contains general configuration information for all
    225 Alpha-based platforms using ELF.  In addition to reading this
    226 section, please read all other sections that match your target.
    227 </p>
    228 <p>We require binutils 2.11.2 or newer.
    229 Previous binutils releases had a number of problems with DWARF 2
    230 debugging information, not the least of which is incorrect linking of
    231 shared libraries.
    232 </p>
    233 <hr />
    234 <a name="amd64-x-solaris210"></a><a name="amd64-*-solaris2_002e1_005b0-9_005d*"></a>
    235 <h3 class="heading">amd64-*-solaris2.1[0-9]*</h3>
    236 <p>This is a synonym for &lsquo;<samp>x86_64-*-solaris2.1[0-9]*</samp>&rsquo;.
    237 </p>
    238 <hr />
    239 <a name="amdgcn-unknown-amdhsa"></a><a name="amdgcn-unknown-amdhsa-1"></a>
    240 <h3 class="heading">amdgcn-unknown-amdhsa</h3>
    241 <p>AMD GCN GPU target.
    242 </p>
    243 <p>Instead of GNU Binutils, you will need to install LLVM 6, or later, and copy
    244 <samp>bin/llvm-mc</samp> to <samp>amdgcn-unknown-amdhsa/bin/as</samp>,
    245 <samp>bin/lld</samp> to <samp>amdgcn-unknown-amdhsa/bin/ld</samp>,
    246 <samp>bin/llvm-nm</samp> to <samp>amdgcn-unknown-amdhsa/bin/nm</samp>, and
    247 <samp>bin/llvm-ar</samp> to both <samp>bin/amdgcn-unknown-amdhsa-ar</samp> and
    248 <samp>bin/amdgcn-unknown-amdhsa-ranlib</samp>.
    249 </p>
    250 <p>Use Newlib (2019-01-16, or newer).
    251 </p>
    252 <p>To run the binaries, install the HSA Runtime from the
    253 <a href="https://rocm.github.io">ROCm Platform</a>, and use
    254 <samp>libexec/gcc/amdhsa-unknown-amdhsa/<var>version</var>/gcn-run</samp> to launch them
    255 on the GPU.
    256 </p>
    257 <hr />
    258 <a name="arc-x-elf32"></a><a name="arc-*-elf32"></a>
    259 <h3 class="heading">arc-*-elf32</h3>
    260 
    261 <p>Use &lsquo;<samp>configure --target=arc-elf32 --with-cpu=<var>cpu</var> --enable-languages=&quot;c,c++&quot;</samp>&rsquo;
    262 to configure GCC, with <var>cpu</var> being one of &lsquo;<samp>arc600</samp>&rsquo;, &lsquo;<samp>arc601</samp>&rsquo;,
    263 or &lsquo;<samp>arc700</samp>&rsquo;.
    264 </p>
    265 <hr />
    266 <a name="arc-linux-uclibc"></a><a name="arc-linux-uclibc-1"></a>
    267 <h3 class="heading">arc-linux-uclibc</h3>
    268 
    269 <p>Use &lsquo;<samp>configure --target=arc-linux-uclibc --with-cpu=arc700 --enable-languages=&quot;c,c++&quot;</samp>&rsquo; to configure GCC.
    270 </p>
    271 <hr />
    272 <a name="arm-x-eabi"></a><a name="arm-*-eabi"></a>
    273 <h3 class="heading">arm-*-eabi</h3>
    274 <p>ARM-family processors.
    275 </p>
    276 <p>Building the Ada frontend commonly fails (an infinite loop executing
    277 <code>xsinfo</code>) if the host compiler is GNAT 4.8.  Host compilers built from the
    278 GNAT 4.6, 4.9 or 5 release branches are known to succeed.
    279 </p>
    280 <hr />
    281 <a name="avr"></a><a name="avr-1"></a>
    282 <h3 class="heading">avr</h3>
    283 <p>ATMEL AVR-family micro controllers.  These are used in embedded
    284 applications.  There are no standard Unix configurations.
    285 See &ldquo;AVR Options&rdquo; in the main manual
    286 for the list of supported MCU types.
    287 </p>
    288 <p>Use &lsquo;<samp>configure --target=avr --enable-languages=&quot;c&quot;</samp>&rsquo; to configure GCC.
    289 </p>
    290 <p>Further installation notes and other useful information about AVR tools
    291 can also be obtained from:
    292 </p>
    293 <ul>
    294 <li> <a href="http://www.nongnu.org/avr/">http://www.nongnu.org/avr/</a>
    295 </li><li> <a href="http://www.amelek.gda.pl/avr/">http://www.amelek.gda.pl/avr/</a>
    296 </li></ul>
    297 
    298 <p>The following error:
    299 </p><div class="smallexample">
    300 <pre class="smallexample">Error: register required
    301 </pre></div>
    302 
    303 <p>indicates that you should upgrade to a newer version of the binutils.
    304 </p>
    305 <hr />
    306 <a name="bfin"></a><a name="Blackfin"></a>
    307 <h3 class="heading">Blackfin</h3>
    308 <p>The Blackfin processor, an Analog Devices DSP.
    309 See &ldquo;Blackfin Options&rdquo; in the main manual
    310 </p>
    311 <p>More information, and a version of binutils with support for this processor,
    312 is available at <a href="https://blackfin.uclinux.org">https://blackfin.uclinux.org</a>
    313 </p>
    314 <hr />
    315 <a name="cr16"></a><a name="CR16"></a>
    316 <h3 class="heading">CR16</h3>
    317 <p>The CR16 CompactRISC architecture is a 16-bit architecture. This
    318 architecture is used in embedded applications.
    319 </p>
    320 
    321 <p>See &ldquo;CR16 Options&rdquo; in the main manual for a list of CR16-specific options.
    322 </p>
    323 <p>Use &lsquo;<samp>configure --target=cr16-elf --enable-languages=c,c++</samp>&rsquo; to configure
    324 GCC&nbsp;for building a CR16 elf cross-compiler.
    325 </p>
    326 <p>Use &lsquo;<samp>configure --target=cr16-uclinux --enable-languages=c,c++</samp>&rsquo; to
    327 configure GCC&nbsp;for building a CR16 uclinux cross-compiler.
    328 </p>
    329 <hr />
    330 <a name="cris"></a><a name="CRIS"></a>
    331 <h3 class="heading">CRIS</h3>
    332 <p>CRIS is the CPU architecture in Axis Communications ETRAX system-on-a-chip
    333 series.  These are used in embedded applications.
    334 </p>
    335 <p>See &ldquo;CRIS Options&rdquo; in the main manual
    336 for a list of CRIS-specific options.
    337 </p>
    338 <p>There are a few different CRIS targets:
    339 </p><dl compact="compact">
    340 <dt><code>cris-axis-elf</code></dt>
    341 <dd><p>Mainly for monolithic embedded systems.  Includes a multilib for the
    342 &lsquo;<samp>v10</samp>&rsquo; core used in &lsquo;<samp>ETRAX 100 LX</samp>&rsquo;.
    343 </p></dd>
    344 <dt><code>cris-axis-linux-gnu</code></dt>
    345 <dd><p>A GNU/Linux port for the CRIS architecture, currently targeting
    346 &lsquo;<samp>ETRAX 100 LX</samp>&rsquo; by default.
    347 </p></dd>
    348 </dl>
    349 
    350 <p>Pre-packaged tools can be obtained from
    351 <a href="ftp://ftp.axis.com/pub/axis/tools/cris/compiler-kit/">ftp://ftp.axis.com/pub/axis/tools/cris/compiler-kit/</a>.  More
    352 information about this platform is available at
    353 <a href="http://developer.axis.com/">http://developer.axis.com/</a>.
    354 </p>
    355 <hr />
    356 <a name="dos"></a><a name="DOS"></a>
    357 <h3 class="heading">DOS</h3>
    358 <p>Please have a look at the <a href="binaries.html">binaries page</a>.
    359 </p>
    360 <p>You cannot install GCC by itself on MSDOS; it will not compile under
    361 any MSDOS compiler except itself.  You need to get the complete
    362 compilation package DJGPP, which includes binaries as well as sources,
    363 and includes all the necessary compilation tools and libraries.
    364 </p>
    365 <hr />
    366 <a name="epiphany-x-elf"></a><a name="epiphany-*-elf"></a>
    367 <h3 class="heading">epiphany-*-elf</h3>
    368 <p>Adapteva Epiphany.
    369 This configuration is intended for embedded systems.
    370 </p>
    371 <hr />
    372 <a name="x-x-freebsd"></a><a name="g_t*-*-freebsd*"></a>
    373 <h3 class="heading">*-*-freebsd*</h3>
    374 <p>Support for FreeBSD 1 was discontinued in GCC 3.2.  Support for
    375 FreeBSD 2 (and any mutant a.out variants of FreeBSD 3) was
    376 discontinued in GCC 4.0.
    377 </p>
    378 <p>In order to better utilize FreeBSD base system functionality and match
    379 the configuration of the system compiler, GCC 4.5 and above as well as
    380 GCC 4.4 past 2010-06-20 leverage SSP support in libc (which is present
    381 on FreeBSD 7 or later) and the use of <code>__cxa_atexit</code> by default
    382 (on FreeBSD 6 or later).  The use of <code>dl_iterate_phdr</code> inside
    383 <samp>libgcc_s.so.1</samp> and boehm-gc (on FreeBSD 7 or later) is enabled
    384 by GCC 4.5 and above.
    385 </p>
    386 <p>We support FreeBSD using the ELF file format with DWARF 2 debugging
    387 for all CPU architectures.  You may use <samp>-gstabs</samp> instead of
    388 <samp>-g</samp>, if you really want the old debugging format.  There are
    389 no known issues with mixing object files and libraries with different
    390 debugging formats.  Otherwise, this release of GCC should now match
    391 more of the configuration used in the stock FreeBSD configuration of
    392 GCC.  In particular, <samp>--enable-threads</samp> is now configured by
    393 default.  However, as a general user, do not attempt to replace the
    394 system compiler with this release.  Known to bootstrap and check with
    395 good results on FreeBSD 7.2-STABLE.  In the past, known to bootstrap
    396 and check with good results on FreeBSD 3.0, 3.4, 4.0, 4.2, 4.3, 4.4,
    397 4.5, 4.8, 4.9 and 5-CURRENT.
    398 </p>
    399 <p>The version of binutils installed in <samp>/usr/bin</samp> probably works
    400 with this release of GCC.  Bootstrapping against the latest GNU
    401 binutils and/or the version found in <samp>/usr/ports/devel/binutils</samp> has
    402 been known to enable additional features and improve overall testsuite
    403 results.  However, it is currently known that boehm-gc may not configure
    404 properly on FreeBSD prior to the FreeBSD 7.0 release with GNU binutils
    405 after 2.16.1.
    406 </p>
    407 <hr />
    408 <a name="ft32-x-elf"></a><a name="ft32-*-elf"></a>
    409 <h3 class="heading">ft32-*-elf</h3>
    410 <p>The FT32 processor.
    411 This configuration is intended for embedded systems.
    412 </p>
    413 <hr />
    414 <a name="h8300-hms"></a><a name="h8300-hms-1"></a>
    415 <h3 class="heading">h8300-hms</h3>
    416 <p>Renesas H8/300 series of processors.
    417 </p>
    418 <p>Please have a look at the <a href="binaries.html">binaries page</a>.
    419 </p>
    420 <p>The calling convention and structure layout has changed in release 2.6.
    421 All code must be recompiled.  The calling convention now passes the
    422 first three arguments in function calls in registers.  Structures are no
    423 longer a multiple of 2 bytes.
    424 </p>
    425 <hr />
    426 <a name="hppa-hp-hpux"></a><a name="hppa*-hp-hpux*"></a>
    427 <h3 class="heading">hppa*-hp-hpux*</h3>
    428 <p>Support for HP-UX version 9 and older was discontinued in GCC 3.4.
    429 </p>
    430 <p>We require using gas/binutils on all hppa platforms.  Version 2.19 or
    431 later is recommended.
    432 </p>
    433 <p>It may be helpful to configure GCC with the
    434 <a href="./configure.html#with-gnu-as"><samp>--with-gnu-as</samp></a> and
    435 <samp>--with-as=&hellip;</samp> options to ensure that GCC can find GAS.
    436 </p>
    437 <p>The HP assembler should not be used with GCC.  It is rarely tested and may
    438 not work.  It shouldn&rsquo;t be used with any languages other than C due to its
    439 many limitations.
    440 </p>
    441 <p>Specifically, <samp>-g</samp> does not work (HP-UX uses a peculiar debugging
    442 format which GCC does not know about).  It also inserts timestamps
    443 into each object file it creates, causing the 3-stage comparison test to
    444 fail during a bootstrap.  You should be able to continue by saying
    445 &lsquo;<samp>make all-host all-target</samp>&rsquo; after getting the failure from &lsquo;<samp>make</samp>&rsquo;.
    446 </p>
    447 <p>Various GCC features are not supported.  For example, it does not support weak
    448 symbols or alias definitions.  As a result, explicit template instantiations
    449 are required when using C++.  This makes it difficult if not impossible to
    450 build many C++ applications.
    451 </p>
    452 <p>There are two default scheduling models for instructions.  These are
    453 PROCESSOR_7100LC and PROCESSOR_8000.  They are selected from the pa-risc
    454 architecture specified for the target machine when configuring.
    455 PROCESSOR_8000 is the default.  PROCESSOR_7100LC is selected when
    456 the target is a &lsquo;<samp>hppa1*</samp>&rsquo; machine.
    457 </p>
    458 <p>The PROCESSOR_8000 model is not well suited to older processors.  Thus,
    459 it is important to completely specify the machine architecture when
    460 configuring if you want a model other than PROCESSOR_8000.  The macro
    461 TARGET_SCHED_DEFAULT can be defined in BOOT_CFLAGS if a different
    462 default scheduling model is desired.
    463 </p>
    464 <p>As of GCC 4.0, GCC uses the UNIX 95 namespace for HP-UX 10.10
    465 through 11.00, and the UNIX 98 namespace for HP-UX 11.11 and later.
    466 This namespace change might cause problems when bootstrapping with
    467 an earlier version of GCC or the HP compiler as essentially the same
    468 namespace is required for an entire build.  This problem can be avoided
    469 in a number of ways.  With HP cc, <code>UNIX_STD</code> can be set to &lsquo;<samp>95</samp>&rsquo;
    470 or &lsquo;<samp>98</samp>&rsquo;.  Another way is to add an appropriate set of predefines
    471 to <code>CC</code>.  The description for the <samp>munix=</samp> option contains
    472 a list of the predefines used with each standard.
    473 </p>
    474 <p>More specific information to &lsquo;<samp>hppa*-hp-hpux*</samp>&rsquo; targets follows.
    475 </p>
    476 <hr />
    477 <a name="hppa-hp-hpux10"></a><a name="hppa*-hp-hpux10"></a>
    478 <h3 class="heading">hppa*-hp-hpux10</h3>
    479 <p>For hpux10.20, we <em>highly</em> recommend you pick up the latest sed patch
    480 <code>PHCO_19798</code> from HP.
    481 </p>
    482 <p>The C++ ABI has changed incompatibly in GCC 4.0.  COMDAT subspaces are
    483 used for one-only code and data.  This resolves many of the previous
    484 problems in using C++ on this target.  However, the ABI is not compatible
    485 with the one implemented under HP-UX 11 using secondary definitions.
    486 </p>
    487 <hr />
    488 <a name="hppa-hp-hpux11"></a><a name="hppa*-hp-hpux11"></a>
    489 <h3 class="heading">hppa*-hp-hpux11</h3>
    490 <p>GCC 3.0 and up support HP-UX 11.  GCC 2.95.x is not supported and cannot
    491 be used to compile GCC 3.0 and up.
    492 </p>
    493 <p>The libffi library haven&rsquo;t been ported to 64-bit HP-UX&nbsp;and doesn&rsquo;t build.
    494 </p>
    495 <p>Refer to <a href="binaries.html">binaries</a> for information about obtaining
    496 precompiled GCC binaries for HP-UX.  Precompiled binaries must be obtained
    497 to build the Ada language as it cannot be bootstrapped using C.  Ada is
    498 only available for the 32-bit PA-RISC runtime.
    499 </p>
    500 <p>Starting with GCC 3.4 an ISO C compiler is required to bootstrap.  The
    501 bundled compiler supports only traditional C; you will need either HP&rsquo;s
    502 unbundled compiler, or a binary distribution of GCC.
    503 </p>
    504 <p>It is possible to build GCC 3.3 starting with the bundled HP compiler,
    505 but the process requires several steps.  GCC 3.3 can then be used to
    506 build later versions.
    507 </p>
    508 <p>There are several possible approaches to building the distribution.
    509 Binutils can be built first using the HP tools.  Then, the GCC
    510 distribution can be built.  The second approach is to build GCC
    511 first using the HP tools, then build binutils, then rebuild GCC.
    512 There have been problems with various binary distributions, so it
    513 is best not to start from a binary distribution.
    514 </p>
    515 <p>On 64-bit capable systems, there are two distinct targets.  Different
    516 installation prefixes must be used if both are to be installed on
    517 the same system.  The &lsquo;<samp>hppa[1-2]*-hp-hpux11*</samp>&rsquo; target generates code
    518 for the 32-bit PA-RISC runtime architecture and uses the HP linker.
    519 The &lsquo;<samp>hppa64-hp-hpux11*</samp>&rsquo; target generates 64-bit code for the
    520 PA-RISC 2.0 architecture.
    521 </p>
    522 <p>The script config.guess now selects the target type based on the compiler
    523 detected during configuration.  You must define <code>PATH</code> or <code>CC</code> so
    524 that configure finds an appropriate compiler for the initial bootstrap.
    525 When <code>CC</code> is used, the definition should contain the options that are
    526 needed whenever <code>CC</code> is used.
    527 </p>
    528 <p>Specifically, options that determine the runtime architecture must be
    529 in <code>CC</code> to correctly select the target for the build.  It is also
    530 convenient to place many other compiler options in <code>CC</code>.  For example,
    531 <code>CC=&quot;cc -Ac +DA2.0W -Wp,-H16376 -D_CLASSIC_TYPES -D_HPUX_SOURCE&quot;</code>
    532 can be used to bootstrap the GCC 3.3 branch with the HP compiler in
    533 64-bit K&amp;R/bundled mode.  The <samp>+DA2.0W</samp> option will result in
    534 the automatic selection of the &lsquo;<samp>hppa64-hp-hpux11*</samp>&rsquo; target.  The
    535 macro definition table of cpp needs to be increased for a successful
    536 build with the HP compiler.  _CLASSIC_TYPES and _HPUX_SOURCE need to
    537 be defined when building with the bundled compiler, or when using the
    538 <samp>-Ac</samp> option.  These defines aren&rsquo;t necessary with <samp>-Ae</samp>.
    539 </p>
    540 <p>It is best to explicitly configure the &lsquo;<samp>hppa64-hp-hpux11*</samp>&rsquo; target
    541 with the <samp>--with-ld=&hellip;</samp> option.  This overrides the standard
    542 search for ld.  The two linkers supported on this target require different
    543 commands.  The default linker is determined during configuration.  As a
    544 result, it&rsquo;s not possible to switch linkers in the middle of a GCC build.
    545 This has been reported to sometimes occur in unified builds of binutils
    546 and GCC.
    547 </p>
    548 <p>A recent linker patch must be installed for the correct operation of
    549 GCC 3.3 and later.  <code>PHSS_26559</code> and <code>PHSS_24304</code> are the
    550 oldest linker patches that are known to work.  They are for HP-UX
    551 11.00 and 11.11, respectively.  <code>PHSS_24303</code>, the companion to
    552 <code>PHSS_24304</code>, might be usable but it hasn&rsquo;t been tested.  These
    553 patches have been superseded.  Consult the HP patch database to obtain
    554 the currently recommended linker patch for your system.
    555 </p>
    556 <p>The patches are necessary for the support of weak symbols on the
    557 32-bit port, and for the running of initializers and finalizers.  Weak
    558 symbols are implemented using SOM secondary definition symbols.  Prior
    559 to HP-UX 11, there are bugs in the linker support for secondary symbols.
    560 The patches correct a problem of linker core dumps creating shared
    561 libraries containing secondary symbols, as well as various other
    562 linking issues involving secondary symbols.
    563 </p>
    564 <p>GCC 3.3 uses the ELF DT_INIT_ARRAY and DT_FINI_ARRAY capabilities to
    565 run initializers and finalizers on the 64-bit port.  The 32-bit port
    566 uses the linker <samp>+init</samp> and <samp>+fini</samp> options for the same
    567 purpose.  The patches correct various problems with the +init/+fini
    568 options, including program core dumps.  Binutils 2.14 corrects a
    569 problem on the 64-bit port resulting from HP&rsquo;s non-standard use of
    570 the .init and .fini sections for array initializers and finalizers.
    571 </p>
    572 <p>Although the HP and GNU linkers are both supported for the
    573 &lsquo;<samp>hppa64-hp-hpux11*</samp>&rsquo; target, it is strongly recommended that the
    574 HP linker be used for link editing on this target.
    575 </p>
    576 <p>At this time, the GNU linker does not support the creation of long
    577 branch stubs.  As a result, it cannot successfully link binaries
    578 containing branch offsets larger than 8 megabytes.  In addition,
    579 there are problems linking shared libraries, linking executables
    580 with <samp>-static</samp>, and with dwarf2 unwind and exception support.
    581 It also doesn&rsquo;t provide stubs for internal calls to global functions
    582 in shared libraries, so these calls cannot be overloaded.
    583 </p>
    584 <p>The HP dynamic loader does not support GNU symbol versioning, so symbol
    585 versioning is not supported.  It may be necessary to disable symbol
    586 versioning with <samp>--disable-symvers</samp> when using GNU ld.
    587 </p>
    588 <p>POSIX threads are the default.  The optional DCE thread library is not
    589 supported, so <samp>--enable-threads=dce</samp> does not work.
    590 </p>
    591 <hr />
    592 <a name="x-x-linux-gnu"></a><a name="g_t*-*-linux-gnu"></a>
    593 <h3 class="heading">*-*-linux-gnu</h3>
    594 <p>Versions of libstdc++-v3 starting with 3.2.1 require bug fixes present
    595 in glibc 2.2.5 and later.  More information is available in the
    596 libstdc++-v3 documentation.
    597 </p>
    598 <hr />
    599 <a name="ix86-x-linux"></a><a name="i_003f86-*-linux*"></a>
    600 <h3 class="heading">i?86-*-linux*</h3>
    601 <p>As of GCC 3.3, binutils 2.13.1 or later is required for this platform.
    602 See <a href="http://gcc.gnu.org/PR10877">bug 10877</a> for more information.
    603 </p>
    604 <p>If you receive Signal 11 errors when building on GNU/Linux, then it is
    605 possible you have a hardware problem.  Further information on this can be
    606 found on <a href="http://www.bitwizard.nl/sig11/">www.bitwizard.nl</a>.
    607 </p>
    608 <hr />
    609 <a name="ix86-x-solaris210"></a><a name="i_003f86-*-solaris2_002e10"></a>
    610 <h3 class="heading">i?86-*-solaris2.10</h3>
    611 <p>Use this for Solaris 10 or later on x86 and x86-64 systems.  Starting
    612 with GCC 4.7, there is also a 64-bit &lsquo;<samp>amd64-*-solaris2.1[0-9]*</samp>&rsquo; or
    613 &lsquo;<samp>x86_64-*-solaris2.1[0-9]*</samp>&rsquo; configuration that corresponds to
    614 &lsquo;<samp>sparcv9-sun-solaris2*</samp>&rsquo;.
    615 </p>
    616 <p>It is recommended that you configure GCC to use the GNU assembler.  The
    617 versions included in Solaris 10, from GNU binutils 2.15 (in
    618 <samp>/usr/sfw/bin/gas</samp>), and Solaris 11, from GNU binutils 2.19 or
    619 newer (also available as <samp>/usr/bin/gas</samp> and
    620 <samp>/usr/gnu/bin/as</samp>), work fine.  The current version, from GNU
    621 binutils 2.29, is known to work, but the version from GNU binutils 2.26
    622 must be avoided.  Recent versions of the Solaris assembler in
    623 <samp>/usr/ccs/bin/as</samp> work almost as well, though.
    624 </p>
    625 <p>For linking, the Solaris linker, is preferred.  If you want to use the GNU
    626 linker instead, note that due to a packaging bug the version in Solaris
    627 10, from GNU binutils 2.15 (in <samp>/usr/sfw/bin/gld</samp>), cannot be used,
    628 while the version in Solaris 11, from GNU binutils 2.19 or newer (also
    629 in <samp>/usr/gnu/bin/ld</samp> and <samp>/usr/bin/gld</samp>), works, as does the
    630 latest version, from GNU binutils 2.29.
    631 </p>
    632 <p>To use GNU <code>as</code>, configure with the options
    633 <samp>--with-gnu-as --with-as=/usr/sfw/bin/gas</samp>.  It may be necessary
    634 to configure with <samp>--without-gnu-ld --with-ld=/usr/ccs/bin/ld</samp> to
    635 guarantee use of Sun <code>ld</code>.
    636 </p>
    637 <hr />
    638 <a name="ia64-x-linux"></a><a name="ia64-*-linux"></a>
    639 <h3 class="heading">ia64-*-linux</h3>
    640 <p>IA-64 processor (also known as IPF, or Itanium Processor Family)
    641 running GNU/Linux.
    642 </p>
    643 <p>If you are using the installed system libunwind library with
    644 <samp>--with-system-libunwind</samp>, then you must use libunwind 0.98 or
    645 later.
    646 </p>
    647 <p>None of the following versions of GCC has an ABI that is compatible
    648 with any of the other versions in this list, with the exception that
    649 Red Hat 2.96 and Trillian 000171 are compatible with each other:
    650 3.1, 3.0.2, 3.0.1, 3.0, Red Hat 2.96, and Trillian 000717.
    651 This primarily affects C++ programs and programs that create shared libraries.
    652 GCC 3.1 or later is recommended for compiling linux, the kernel.
    653 As of version 3.1 GCC is believed to be fully ABI compliant, and hence no
    654 more major ABI changes are expected.
    655 </p>
    656 <hr />
    657 <a name="ia64-x-hpux"></a><a name="ia64-*-hpux*"></a>
    658 <h3 class="heading">ia64-*-hpux*</h3>
    659 <p>Building GCC on this target requires the GNU Assembler.  The bundled HP
    660 assembler will not work.  To prevent GCC from using the wrong assembler,
    661 the option <samp>--with-gnu-as</samp> may be necessary.
    662 </p>
    663 <p>The GCC libunwind library has not been ported to HPUX.  This means that for
    664 GCC versions 3.2.3 and earlier, <samp>--enable-libunwind-exceptions</samp>
    665 is required to build GCC.  For GCC 3.3 and later, this is the default.
    666 For gcc 3.4.3 and later, <samp>--enable-libunwind-exceptions</samp> is
    667 removed and the system libunwind library will always be used.
    668 </p>
    669 <hr />
    670 <!-- rs6000-ibm-aix*, powerpc-ibm-aix* -->
    671 <a name="x-ibm-aix"></a><a name="g_t*-ibm-aix*"></a>
    672 <h3 class="heading">*-ibm-aix*</h3>
    673 <p>Support for AIX version 3 and older was discontinued in GCC 3.4.
    674 Support for AIX version 4.2 and older was discontinued in GCC 4.5.
    675 </p>
    676 <p>&ldquo;out of memory&rdquo; bootstrap failures may indicate a problem with
    677 process resource limits (ulimit).  Hard limits are configured in the
    678 <samp>/etc/security/limits</samp> system configuration file.
    679 </p>
    680 <p>GCC 4.9 and above require a C++ compiler for bootstrap.  IBM VAC++ / xlC
    681 cannot bootstrap GCC.  xlc can bootstrap an older version of GCC and
    682 G++ can bootstrap recent releases of GCC.
    683 </p>
    684 <p>GCC can bootstrap with recent versions of IBM XLC, but bootstrapping
    685 with an earlier release of GCC is recommended.  Bootstrapping with XLC
    686 requires a larger data segment, which can be enabled through the
    687 <var>LDR_CNTRL</var> environment variable, e.g.,
    688 </p>
    689 <div class="smallexample">
    690 <pre class="smallexample">% LDR_CNTRL=MAXDATA=0x50000000
    691 % export LDR_CNTRL
    692 </pre></div>
    693 
    694 <p>One can start with a pre-compiled version of GCC to build from
    695 sources.  One may delete GCC&rsquo;s &ldquo;fixed&rdquo; header files when starting
    696 with a version of GCC built for an earlier release of AIX.
    697 </p>
    698 <p>To speed up the configuration phases of bootstrapping and installing GCC,
    699 one may use GNU Bash instead of AIX <code>/bin/sh</code>, e.g.,
    700 </p>
    701 <div class="smallexample">
    702 <pre class="smallexample">% CONFIG_SHELL=/opt/freeware/bin/bash
    703 % export CONFIG_SHELL
    704 </pre></div>
    705 
    706 <p>and then proceed as described in <a href="build.html">the build
    707 instructions</a>, where we strongly recommend specifying an absolute path
    708 to invoke <var>srcdir</var>/configure.
    709 </p>
    710 <p>Because GCC on AIX is built as a 32-bit executable by default,
    711 (although it can generate 64-bit programs) the GMP and MPFR libraries
    712 required by gfortran must be 32-bit libraries.  Building GMP and MPFR
    713 as static archive libraries works better than shared libraries.
    714 </p>
    715 <p>Errors involving <code>alloca</code> when building GCC generally are due
    716 to an incorrect definition of <code>CC</code> in the Makefile or mixing files
    717 compiled with the native C compiler and GCC.  During the stage1 phase of
    718 the build, the native AIX compiler <strong>must</strong> be invoked as <code>cc</code>
    719 (not <code>xlc</code>).  Once <code>configure</code> has been informed of
    720 <code>xlc</code>, one needs to use &lsquo;<samp>make distclean</samp>&rsquo; to remove the
    721 configure cache files and ensure that <code>CC</code> environment variable
    722 does not provide a definition that will confuse <code>configure</code>.
    723 If this error occurs during stage2 or later, then the problem most likely
    724 is the version of Make (see above).
    725 </p>
    726 <p>The native <code>as</code> and <code>ld</code> are recommended for
    727 bootstrapping on AIX.  The GNU Assembler, GNU Linker, and GNU
    728 Binutils version 2.20 is the minimum level that supports bootstrap on
    729 AIX 5.  The GNU Assembler has not been updated to support AIX 6&nbsp;or
    730 AIX 7.  The native AIX tools do interoperate with GCC.
    731 </p>
    732 <p>AIX 7.1 added partial support for DWARF debugging, but full support
    733 requires AIX 7.1 TL03 SP7 that supports additional DWARF sections and
    734 fixes a bug in the assembler.  AIX 7.1 TL03 SP5 distributed a version
    735 of libm.a missing important symbols; a fix for IV77796 will be
    736 included in SP6.
    737 </p>
    738 <p>AIX 5.3 TL10, AIX 6.1 TL05 and AIX 7.1 TL00 introduced an AIX
    739 assembler change that sometimes produces corrupt assembly files
    740 causing AIX linker errors.  The bug breaks GCC bootstrap on AIX and
    741 can cause compilation failures with existing GCC installations.  An
    742 AIX iFix for AIX 5.3 is available (APAR IZ98385 for AIX 5.3 TL10, APAR
    743 IZ98477 for AIX 5.3 TL11 and IZ98134 for AIX 5.3 TL12). AIX 5.3 TL11 SP8,
    744 AIX 5.3 TL12 SP5, AIX 6.1 TL04 SP11, AIX 6.1 TL05 SP7, AIX 6.1 TL06 SP6,
    745 AIX 6.1 TL07 and AIX 7.1 TL01 should include the fix.
    746 </p>
    747 <p>Building <samp>libstdc++.a</samp> requires a fix for an AIX Assembler bug
    748 APAR IY26685 (AIX 4.3) or APAR IY25528 (AIX 5.1).  It also requires a
    749 fix for another AIX Assembler bug and a co-dependent AIX Archiver fix
    750 referenced as APAR IY53606 (AIX 5.2) or as APAR IY54774 (AIX 5.1)
    751 </p>
    752 <a name="TransferAixShobj"></a><p>&lsquo;<samp>libstdc++</samp>&rsquo; in GCC 3.4 increments the major version number of the
    753 shared object and GCC installation places the <samp>libstdc++.a</samp>
    754 shared library in a common location which will overwrite the and GCC
    755 3.3 version of the shared library.  Applications either need to be
    756 re-linked against the new shared library or the GCC 3.1 and GCC 3.3
    757 versions of the &lsquo;<samp>libstdc++</samp>&rsquo; shared object needs to be available
    758 to the AIX runtime loader.  The GCC 3.1 &lsquo;<samp>libstdc++.so.4</samp>&rsquo;, if
    759 present, and GCC 3.3 &lsquo;<samp>libstdc++.so.5</samp>&rsquo; shared objects can be
    760 installed for runtime dynamic loading using the following steps to set
    761 the &lsquo;<samp>F_LOADONLY</samp>&rsquo; flag in the shared object for <em>each</em>
    762 multilib <samp>libstdc++.a</samp> installed:
    763 </p>
    764 <p>Extract the shared objects from the currently installed
    765 <samp>libstdc++.a</samp> archive:
    766 </p><div class="smallexample">
    767 <pre class="smallexample">% ar -x libstdc++.a libstdc++.so.4 libstdc++.so.5
    768 </pre></div>
    769 
    770 <p>Enable the &lsquo;<samp>F_LOADONLY</samp>&rsquo; flag so that the shared object will be
    771 available for runtime dynamic loading, but not linking:
    772 </p><div class="smallexample">
    773 <pre class="smallexample">% strip -e libstdc++.so.4 libstdc++.so.5
    774 </pre></div>
    775 
    776 <p>Archive the runtime-only shared object in the GCC 3.4
    777 <samp>libstdc++.a</samp> archive:
    778 </p><div class="smallexample">
    779 <pre class="smallexample">% ar -q libstdc++.a libstdc++.so.4 libstdc++.so.5
    780 </pre></div>
    781 
    782 <p>Eventually, the
    783 <a href="./configure.html#WithAixSoname"><samp>--with-aix-soname=svr4</samp></a>
    784 configure option may drop the need for this procedure for libraries that
    785 support it.
    786 </p>
    787 <p>Linking executables and shared libraries may produce warnings of
    788 duplicate symbols.  The assembly files generated by GCC for AIX always
    789 have included multiple symbol definitions for certain global variable
    790 and function declarations in the original program.  The warnings should
    791 not prevent the linker from producing a correct library or runnable
    792 executable.
    793 </p>
    794 <p>AIX 4.3 utilizes a &ldquo;large format&rdquo; archive to support both 32-bit and
    795 64-bit object modules.  The routines provided in AIX 4.3.0 and AIX 4.3.1
    796 to parse archive libraries did not handle the new format correctly.
    797 These routines are used by GCC and result in error messages during
    798 linking such as &ldquo;not a COFF file&rdquo;.  The version of the routines shipped
    799 with AIX 4.3.1 should work for a 32-bit environment.  The <samp>-g</samp>
    800 option of the archive command may be used to create archives of 32-bit
    801 objects using the original &ldquo;small format&rdquo;.  A correct version of the
    802 routines is shipped with AIX 4.3.2 and above.
    803 </p>
    804 <p>Some versions of the AIX binder (linker) can fail with a relocation
    805 overflow severe error when the <samp>-bbigtoc</samp> option is used to link
    806 GCC-produced object files into an executable that overflows the TOC.  A fix
    807 for APAR IX75823 (OVERFLOW DURING LINK WHEN USING GCC AND -BBIGTOC) is
    808 available from IBM Customer Support and from its
    809 <a href="http://techsupport.services.ibm.com/">techsupport.services.ibm.com</a>
    810 website as PTF U455193.
    811 </p>
    812 <p>The AIX 4.3.2.1 linker (bos.rte.bind_cmds Level 4.3.2.1) will dump core
    813 with a segmentation fault when invoked by any version of GCC.  A fix for
    814 APAR IX87327 is available from IBM Customer Support and from its
    815 <a href="http://techsupport.services.ibm.com/">techsupport.services.ibm.com</a>
    816 website as PTF U461879.  This fix is incorporated in AIX 4.3.3 and above.
    817 </p>
    818 <p>The initial assembler shipped with AIX 4.3.0 generates incorrect object
    819 files.  A fix for APAR IX74254 (64BIT DISASSEMBLED OUTPUT FROM COMPILER FAILS
    820 TO ASSEMBLE/BIND) is available from IBM Customer Support and from its
    821 <a href="http://techsupport.services.ibm.com/">techsupport.services.ibm.com</a>
    822 website as PTF U453956.  This fix is incorporated in AIX 4.3.1 and above.
    823 </p>
    824 <p>AIX provides National Language Support (NLS).  Compilers and assemblers
    825 use NLS to support locale-specific representations of various data
    826 formats including floating-point numbers (e.g., &lsquo;<samp>.</samp>&rsquo;  vs &lsquo;<samp>,</samp>&rsquo; for
    827 separating decimal fractions).  There have been problems reported where
    828 GCC does not produce the same floating-point formats that the assembler
    829 expects.  If one encounters this problem, set the <code>LANG</code>
    830 environment variable to &lsquo;<samp>C</samp>&rsquo; or &lsquo;<samp>En_US</samp>&rsquo;.
    831 </p>
    832 <p>A default can be specified with the <samp>-mcpu=<var>cpu_type</var></samp>
    833 switch and using the configure option <samp>--with-cpu-<var>cpu_type</var></samp>.
    834 </p>
    835 <hr />
    836 <a name="iq2000-x-elf"></a><a name="iq2000-*-elf"></a>
    837 <h3 class="heading">iq2000-*-elf</h3>
    838 <p>Vitesse IQ2000 processors.  These are used in embedded
    839 applications.  There are no standard Unix configurations.
    840 </p>
    841 <hr />
    842 <a name="lm32-x-elf"></a><a name="lm32-*-elf"></a>
    843 <h3 class="heading">lm32-*-elf</h3>
    844 <p>Lattice Mico32 processor.
    845 This configuration is intended for embedded systems.
    846 </p>
    847 <hr />
    848 <a name="lm32-x-uclinux"></a><a name="lm32-*-uclinux"></a>
    849 <h3 class="heading">lm32-*-uclinux</h3>
    850 <p>Lattice Mico32 processor.
    851 This configuration is intended for embedded systems running uClinux.
    852 </p>
    853 <hr />
    854 <a name="m32c-x-elf"></a><a name="m32c-*-elf"></a>
    855 <h3 class="heading">m32c-*-elf</h3>
    856 <p>Renesas M32C processor.
    857 This configuration is intended for embedded systems.
    858 </p>
    859 <hr />
    860 <a name="m32r-x-elf"></a><a name="m32r-*-elf"></a>
    861 <h3 class="heading">m32r-*-elf</h3>
    862 <p>Renesas M32R processor.
    863 This configuration is intended for embedded systems.
    864 </p>
    865 <hr />
    866 <a name="m68k-x-x"></a><a name="m68k-*-*"></a>
    867 <h3 class="heading">m68k-*-*</h3>
    868 <p>By default,
    869 &lsquo;<samp>m68k-*-elf*</samp>&rsquo;, &lsquo;<samp>m68k-*-rtems</samp>&rsquo;,  &lsquo;<samp>m68k-*-uclinux</samp>&rsquo; and
    870 &lsquo;<samp>m68k-*-linux</samp>&rsquo;
    871 build libraries for both M680x0 and ColdFire processors.  If you only
    872 need the M680x0 libraries, you can omit the ColdFire ones by passing
    873 <samp>--with-arch=m68k</samp> to <code>configure</code>.  Alternatively, you
    874 can omit the M680x0 libraries by passing <samp>--with-arch=cf</samp> to
    875 <code>configure</code>.  These targets default to 5206 or 5475 code as
    876 appropriate for the target system when
    877 configured with <samp>--with-arch=cf</samp> and 68020 code otherwise.
    878 </p>
    879 <p>The &lsquo;<samp>m68k-*-netbsd</samp>&rsquo; and
    880 &lsquo;<samp>m68k-*-openbsd</samp>&rsquo; targets also support the <samp>--with-arch</samp>
    881 option.  They will generate ColdFire CFV4e code when configured with
    882 <samp>--with-arch=cf</samp> and 68020 code otherwise.
    883 </p>
    884 <p>You can override the default processors listed above by configuring
    885 with <samp>--with-cpu=<var>target</var></samp>.  This <var>target</var> can either
    886 be a <samp>-mcpu</samp> argument or one of the following values:
    887 &lsquo;<samp>m68000</samp>&rsquo;, &lsquo;<samp>m68010</samp>&rsquo;, &lsquo;<samp>m68020</samp>&rsquo;, &lsquo;<samp>m68030</samp>&rsquo;,
    888 &lsquo;<samp>m68040</samp>&rsquo;, &lsquo;<samp>m68060</samp>&rsquo;, &lsquo;<samp>m68020-40</samp>&rsquo; and &lsquo;<samp>m68020-60</samp>&rsquo;.
    889 </p>
    890 <p>GCC requires at least binutils version 2.17 on these targets.
    891 </p>
    892 <hr />
    893 <a name="m68k-x-uclinux"></a><a name="m68k-*-uclinux"></a>
    894 <h3 class="heading">m68k-*-uclinux</h3>
    895 <p>GCC 4.3 changed the uClinux configuration so that it uses the
    896 &lsquo;<samp>m68k-linux-gnu</samp>&rsquo; ABI rather than the &lsquo;<samp>m68k-elf</samp>&rsquo; ABI.
    897 It also added improved support for C++ and flat shared libraries,
    898 both of which were ABI changes.
    899 </p>
    900 <hr />
    901 <a name="microblaze-x-elf"></a><a name="microblaze-*-elf"></a>
    902 <h3 class="heading">microblaze-*-elf</h3>
    903 <p>Xilinx MicroBlaze processor.
    904 This configuration is intended for embedded systems.
    905 </p>
    906 <hr />
    907 <a name="mips-x-x"></a><a name="mips-*-*"></a>
    908 <h3 class="heading">mips-*-*</h3>
    909 <p>If on a MIPS system you get an error message saying &ldquo;does not have gp
    910 sections for all it&rsquo;s [sic] sectons [sic]&rdquo;, don&rsquo;t worry about it.  This
    911 happens whenever you use GAS with the MIPS linker, but there is not
    912 really anything wrong, and it is okay to use the output file.  You can
    913 stop such warnings by installing the GNU linker.
    914 </p>
    915 <p>It would be nice to extend GAS to produce the gp tables, but they are
    916 optional, and there should not be a warning about their absence.
    917 </p>
    918 <p>The libstdc++ atomic locking routines for MIPS targets requires MIPS II
    919 and later.  A patch went in just after the GCC 3.3 release to
    920 make &lsquo;<samp>mips*-*-*</samp>&rsquo; use the generic implementation instead.  You can also
    921 configure for &lsquo;<samp>mipsel-elf</samp>&rsquo; as a workaround.  The
    922 &lsquo;<samp>mips*-*-linux*</samp>&rsquo; target continues to use the MIPS II routines.  More
    923 work on this is expected in future releases.
    924 </p>
    925 
    926 <p>The built-in <code>__sync_*</code> functions are available on MIPS II and
    927 later systems and others that support the &lsquo;<samp>ll</samp>&rsquo;, &lsquo;<samp>sc</samp>&rsquo; and
    928 &lsquo;<samp>sync</samp>&rsquo; instructions.  This can be overridden by passing
    929 <samp>--with-llsc</samp> or <samp>--without-llsc</samp> when configuring GCC.
    930 Since the Linux kernel emulates these instructions if they are
    931 missing, the default for &lsquo;<samp>mips*-*-linux*</samp>&rsquo; targets is
    932 <samp>--with-llsc</samp>.  The <samp>--with-llsc</samp> and
    933 <samp>--without-llsc</samp> configure options may be overridden at compile
    934 time by passing the <samp>-mllsc</samp> or <samp>-mno-llsc</samp> options to
    935 the compiler.
    936 </p>
    937 <p>MIPS systems check for division by zero (unless
    938 <samp>-mno-check-zero-division</samp> is passed to the compiler) by
    939 generating either a conditional trap or a break instruction.  Using
    940 trap results in smaller code, but is only supported on MIPS II and
    941 later.  Also, some versions of the Linux kernel have a bug that
    942 prevents trap from generating the proper signal (<code>SIGFPE</code>).  To enable
    943 the use of break, use the <samp>--with-divide=breaks</samp>
    944 <code>configure</code> option when configuring GCC.  The default is to
    945 use traps on systems that support them.
    946 </p>
    947 <hr />
    948 <a name="moxie-x-elf"></a><a name="moxie-*-elf"></a>
    949 <h3 class="heading">moxie-*-elf</h3>
    950 <p>The moxie processor.
    951 </p>
    952 <hr />
    953 <a name="msp430-x-elf"></a><a name="msp430-*-elf"></a>
    954 <h3 class="heading">msp430-*-elf</h3>
    955 <p>TI MSP430 processor.
    956 This configuration is intended for embedded systems.
    957 </p>
    958 <hr />
    959 <a name="nds32le-x-elf"></a><a name="nds32le-*-elf"></a>
    960 <h3 class="heading">nds32le-*-elf</h3>
    961 <p>Andes NDS32 target in little endian mode.
    962 </p>
    963 <hr />
    964 <a name="nds32be-x-elf"></a><a name="nds32be-*-elf"></a>
    965 <h3 class="heading">nds32be-*-elf</h3>
    966 <p>Andes NDS32 target in big endian mode.
    967 </p>
    968 <hr />
    969 <a name="nvptx-x-none"></a><a name="nvptx-*-none"></a>
    970 <h3 class="heading">nvptx-*-none</h3>
    971 <p>Nvidia PTX target.
    972 </p>
    973 <p>Instead of GNU binutils, you will need to install
    974 <a href="https://github.com/MentorEmbedded/nvptx-tools/">nvptx-tools</a>.
    975 Tell GCC where to find it:
    976 <samp>--with-build-time-tools=[install-nvptx-tools]/nvptx-none/bin</samp>.
    977 </p>
    978 <p>You will need newlib 3.0 git revision
    979 cd31fbb2aea25f94d7ecedc9db16dfc87ab0c316 or later.  It can be
    980 automatically built together with GCC.  For this, add a symbolic link
    981 to nvptx-newlib&rsquo;s <samp>newlib</samp> directory to the directory containing
    982 the GCC sources.
    983 </p>
    984 <p>Use the <samp>--disable-sjlj-exceptions</samp> and
    985 <samp>--enable-newlib-io-long-long</samp> options when configuring.
    986 </p>
    987 <hr />
    988 <a name="or1k-x-elf"></a><a name="or1k-*-elf"></a>
    989 <h3 class="heading">or1k-*-elf</h3>
    990 <p>The OpenRISC 1000 32-bit processor with delay slots.
    991 This configuration is intended for embedded systems.
    992 </p>
    993 <hr />
    994 <a name="or1k-x-linux"></a><a name="or1k-*-linux"></a>
    995 <h3 class="heading">or1k-*-linux</h3>
    996 <p>The OpenRISC 1000 32-bit processor with delay slots.
    997 </p>
    998 <hr />
    999 <a name="powerpc-x-x"></a><a name="powerpc-*-*"></a>
   1000 <h3 class="heading">powerpc-*-*</h3>
   1001 <p>You can specify a default version for the <samp>-mcpu=<var>cpu_type</var></samp>
   1002 switch by using the configure option <samp>--with-cpu-<var>cpu_type</var></samp>.
   1003 </p>
   1004 <p>You will need GNU binutils 2.15 or newer.
   1005 </p>
   1006 <hr />
   1007 <a name="powerpc-x-darwin"></a><a name="powerpc-*-darwin*"></a>
   1008 <h3 class="heading">powerpc-*-darwin*</h3>
   1009 <p>PowerPC running Darwin (Mac OS X kernel).
   1010 </p>
   1011 <p>Pre-installed versions of Mac OS X may not include any developer tools,
   1012 meaning that you will not be able to build GCC from source.  Tool
   1013 binaries are available at
   1014 <a href="https://opensource.apple.com">https://opensource.apple.com</a>.
   1015 </p>
   1016 <p>This version of GCC requires at least cctools-590.36.  The
   1017 cctools-590.36 package referenced from
   1018 <a href="http://gcc.gnu.org/ml/gcc/2006-03/msg00507.html">http://gcc.gnu.org/ml/gcc/2006-03/msg00507.html</a> will not work
   1019 on systems older than 10.3.9 (aka darwin7.9.0).
   1020 </p>
   1021 <hr />
   1022 <a name="powerpc-x-elf"></a><a name="powerpc-*-elf"></a>
   1023 <h3 class="heading">powerpc-*-elf</h3>
   1024 <p>PowerPC system in big endian mode, running System V.4.
   1025 </p>
   1026 <hr />
   1027 <a name="powerpc-x-linux-gnu"></a><a name="powerpc*-*-linux-gnu*"></a>
   1028 <h3 class="heading">powerpc*-*-linux-gnu*</h3>
   1029 <p>PowerPC system in big endian mode running Linux.
   1030 </p>
   1031 <hr />
   1032 <a name="powerpc-x-netbsd"></a><a name="powerpc-*-netbsd*"></a>
   1033 <h3 class="heading">powerpc-*-netbsd*</h3>
   1034 <p>PowerPC system in big endian mode running NetBSD.
   1035 </p>
   1036 <hr />
   1037 <a name="powerpc-x-eabisim"></a><a name="powerpc-*-eabisim"></a>
   1038 <h3 class="heading">powerpc-*-eabisim</h3>
   1039 <p>Embedded PowerPC system in big endian mode for use in running under the
   1040 PSIM simulator.
   1041 </p>
   1042 <hr />
   1043 <a name="powerpc-x-eabi"></a><a name="powerpc-*-eabi"></a>
   1044 <h3 class="heading">powerpc-*-eabi</h3>
   1045 <p>Embedded PowerPC system in big endian mode.
   1046 </p>
   1047 <hr />
   1048 <a name="powerpcle-x-elf"></a><a name="powerpcle-*-elf"></a>
   1049 <h3 class="heading">powerpcle-*-elf</h3>
   1050 <p>PowerPC system in little endian mode, running System V.4.
   1051 </p>
   1052 <hr />
   1053 <a name="powerpcle-x-eabisim"></a><a name="powerpcle-*-eabisim"></a>
   1054 <h3 class="heading">powerpcle-*-eabisim</h3>
   1055 <p>Embedded PowerPC system in little endian mode for use in running under
   1056 the PSIM simulator.
   1057 </p>
   1058 <hr />
   1059 <a name="powerpcle-x-eabi"></a><a name="powerpcle-*-eabi"></a>
   1060 <h3 class="heading">powerpcle-*-eabi</h3>
   1061 <p>Embedded PowerPC system in little endian mode.
   1062 </p>
   1063 <hr />
   1064 <a name="rl78-x-elf"></a><a name="rl78-*-elf"></a>
   1065 <h3 class="heading">rl78-*-elf</h3>
   1066 <p>The Renesas RL78 processor.
   1067 This configuration is intended for embedded systems.
   1068 </p>
   1069 <hr />
   1070 <a name="riscv32-x-elf"></a><a name="riscv32-*-elf"></a>
   1071 <h3 class="heading">riscv32-*-elf</h3>
   1072 <p>The RISC-V RV32 instruction set.
   1073 This configuration is intended for embedded systems.
   1074 This (and all other RISC-V) targets are supported upstream as of the
   1075 binutils 2.28 release.
   1076 </p>
   1077 <hr />
   1078 <a name="riscv32-x-linux"></a><a name="riscv32-*-linux"></a>
   1079 <h3 class="heading">riscv32-*-linux</h3>
   1080 <p>The RISC-V RV32 instruction set running GNU/Linux.
   1081 This (and all other RISC-V) targets are supported upstream as of the
   1082 binutils 2.28 release.
   1083 </p>
   1084 <hr />
   1085 <a name="riscv64-x-elf"></a><a name="riscv64-*-elf"></a>
   1086 <h3 class="heading">riscv64-*-elf</h3>
   1087 <p>The RISC-V RV64 instruction set.
   1088 This configuration is intended for embedded systems.
   1089 This (and all other RISC-V) targets are supported upstream as of the
   1090 binutils 2.28 release.
   1091 </p>
   1092 <hr />
   1093 <a name="riscv64-x-linux"></a><a name="riscv64-*-linux"></a>
   1094 <h3 class="heading">riscv64-*-linux</h3>
   1095 <p>The RISC-V RV64 instruction set running GNU/Linux.
   1096 This (and all other RISC-V) targets are supported upstream as of the
   1097 binutils 2.28 release.
   1098 </p>
   1099 <hr />
   1100 <a name="rx-x-elf"></a><a name="rx-*-elf"></a>
   1101 <h3 class="heading">rx-*-elf</h3>
   1102 <p>The Renesas RX processor.
   1103 </p>
   1104 <hr />
   1105 <a name="s390-x-linux"></a><a name="s390-*-linux*"></a>
   1106 <h3 class="heading">s390-*-linux*</h3>
   1107 <p>S/390 system running GNU/Linux for S/390.
   1108 </p>
   1109 <hr />
   1110 <a name="s390x-x-linux"></a><a name="s390x-*-linux*"></a>
   1111 <h3 class="heading">s390x-*-linux*</h3>
   1112 <p>zSeries system (64-bit) running GNU/Linux for zSeries.
   1113 </p>
   1114 <hr />
   1115 <a name="s390x-ibm-tpf"></a><a name="s390x-ibm-tpf*"></a>
   1116 <h3 class="heading">s390x-ibm-tpf*</h3>
   1117 <p>zSeries system (64-bit) running TPF.  This platform is
   1118 supported as cross-compilation target only.
   1119 </p>
   1120 <hr />
   1121 <a name="x-x-solaris2"></a><a name="g_t*-*-solaris2*"></a>
   1122 <h3 class="heading">*-*-solaris2*</h3>
   1123 <p>Support for Solaris 10 has been obsoleted in GCC 9, but can still be
   1124 enabled by configuring with <samp>--enable-obsolete</samp>.  Support will be
   1125 removed in GCC 10.  Support for Solaris 9 has been removed in GCC 5.
   1126 Support for Solaris 8 has been removed in GCC 4.8.  Support for Solaris
   1127 7 has been removed in GCC 4.6.
   1128 </p>
   1129 <p>Sun does not ship a C compiler with Solaris 2 before Solaris 10, though
   1130 you can download the Sun Studio compilers for free.  In Solaris 10 and
   1131 11, GCC 3.4.3 is available as <code>/usr/sfw/bin/gcc</code>.  Solaris 11
   1132 also provides GCC 4.5.2, 4.7.3, and 4.8.2 as
   1133 <code>/usr/gcc/4.5/bin/gcc</code> or similar.  Alternatively,
   1134 you can install a pre-built GCC to bootstrap and install GCC.  See the
   1135 <a href="binaries.html">binaries page</a> for details.
   1136 </p>
   1137 <p>The Solaris 2 <code>/bin/sh</code> will often fail to configure
   1138 &lsquo;<samp>libstdc++-v3</samp>&rsquo;or &lsquo;<samp>boehm-gc</samp>&rsquo;.  We therefore recommend using the
   1139 following initial sequence of commands
   1140 </p>
   1141 <div class="smallexample">
   1142 <pre class="smallexample">% CONFIG_SHELL=/bin/ksh
   1143 % export CONFIG_SHELL
   1144 </pre></div>
   1145 
   1146 <p>and proceed as described in <a href="configure.html">the configure instructions</a>.
   1147 In addition we strongly recommend specifying an absolute path to invoke
   1148 <code><var>srcdir</var>/configure</code>.
   1149 </p>
   1150 <p>Solaris 10 comes with a number of optional OS packages.  Some of these
   1151 are needed to use GCC fully, namely <code>SUNWarc</code>,
   1152 <code>SUNWbtool</code>, <code>SUNWesu</code>, <code>SUNWhea</code>, <code>SUNWlibm</code>,
   1153 <code>SUNWsprot</code>, and <code>SUNWtoo</code>.  If you did not install all
   1154 optional packages when installing Solaris 10, you will need to verify that
   1155 the packages that GCC needs are installed.
   1156 To check whether an optional package is installed, use
   1157 the <code>pkginfo</code> command.  To add an optional package, use the
   1158 <code>pkgadd</code> command.  For further details, see the Solaris 10
   1159 documentation.
   1160 </p>
   1161 <p>Starting with Solaris 11, the package management has changed, so you
   1162 need to check for <code>system/header</code>, <code>system/linker</code>, and
   1163 <code>developer/assembler</code> packages.  Checking for and installing
   1164 packages is done with the <code>pkg</code> command now.
   1165 </p>
   1166 <p>Trying to use the linker and other tools in
   1167 <samp>/usr/ucb</samp> to install GCC has been observed to cause trouble.
   1168 For example, the linker may hang indefinitely.  The fix is to remove
   1169 <samp>/usr/ucb</samp> from your <code>PATH</code>.
   1170 </p>
   1171 <p>The build process works more smoothly with the legacy Sun tools so, if you
   1172 have <samp>/usr/xpg4/bin</samp> in your <code>PATH</code>, we recommend that you place
   1173 <samp>/usr/bin</samp> before <samp>/usr/xpg4/bin</samp> for the duration of the build.
   1174 </p>
   1175 <p>We recommend the use of the Solaris assembler or the GNU assembler, in
   1176 conjunction with the Solaris linker.  The GNU <code>as</code>
   1177 versions included in Solaris 10, from GNU binutils 2.15 (in
   1178 <samp>/usr/sfw/bin/gas</samp>), and Solaris 11,
   1179 from GNU binutils 2.19 or newer (also in <samp>/usr/bin/gas</samp> and
   1180 <samp>/usr/gnu/bin/as</samp>), are known to work.
   1181 The current version, from GNU binutils 2.29,
   1182 is known to work as well.  Note that your mileage may vary
   1183 if you use a combination of the GNU tools and the Solaris tools: while the
   1184 combination GNU <code>as</code> + Sun <code>ld</code> should reasonably work,
   1185 the reverse combination Sun <code>as</code> + GNU <code>ld</code> may fail to
   1186 build or cause memory corruption at runtime in some cases for C++ programs.
   1187 GNU <code>ld</code> usually works as well, although the version included in
   1188 Solaris 10 cannot be used due to several bugs.  Again, the current
   1189 version (2.29) is known to work, but generally lacks platform specific
   1190 features, so better stay with Solaris <code>ld</code>.  To use the LTO linker
   1191 plugin (<samp>-fuse-linker-plugin</samp>) with GNU <code>ld</code>, GNU
   1192 binutils <em>must</em> be configured with <samp>--enable-largefile</samp>.
   1193 </p>
   1194 <p>To enable symbol versioning in &lsquo;<samp>libstdc++</samp>&rsquo; with the Solaris linker,
   1195 you need to have any version of GNU <code>c++filt</code>, which is part of
   1196 GNU binutils.  &lsquo;<samp>libstdc++</samp>&rsquo; symbol versioning will be disabled if no
   1197 appropriate version is found.  Solaris <code>c++filt</code> from the Solaris
   1198 Studio compilers does <em>not</em> work.
   1199 </p>
   1200 <p>Sun bug 4927647 sometimes causes random spurious testsuite failures
   1201 related to missing diagnostic output.  This bug doesn&rsquo;t affect GCC
   1202 itself, rather it is a kernel bug triggered by the <code>expect</code>
   1203 program which is used only by the GCC testsuite driver.  When the bug
   1204 causes the <code>expect</code> program to miss anticipated output, extra
   1205 testsuite failures appear.
   1206 </p>
   1207 <hr />
   1208 <a name="sparc-x-x"></a><a name="sparc*-*-*"></a>
   1209 <h3 class="heading">sparc*-*-*</h3>
   1210 <p>This section contains general configuration information for all
   1211 SPARC-based platforms.  In addition to reading this section, please
   1212 read all other sections that match your target.
   1213 </p>
   1214 <p>Newer versions of the GNU Multiple Precision Library (GMP), the MPFR
   1215 library and the MPC library are known to be miscompiled by earlier
   1216 versions of GCC on these platforms.  We therefore recommend the use
   1217 of the exact versions of these libraries listed as minimal versions
   1218 in <a href="prerequisites.html">the prerequisites</a>.
   1219 </p>
   1220 <hr />
   1221 <a name="sparc-sun-solaris2"></a><a name="sparc-sun-solaris2*"></a>
   1222 <h3 class="heading">sparc-sun-solaris2*</h3>
   1223 <p>When GCC is configured to use GNU binutils 2.14 or later, the binaries
   1224 produced are smaller than the ones produced using Sun&rsquo;s native tools;
   1225 this difference is quite significant for binaries containing debugging
   1226 information.
   1227 </p>
   1228 <p>Starting with Solaris 7, the operating system is capable of executing
   1229 64-bit SPARC V9 binaries.  GCC 3.1 and later properly supports
   1230 this; the <samp>-m64</samp> option enables 64-bit code generation.
   1231 However, if all you want is code tuned for the UltraSPARC CPU, you
   1232 should try the <samp>-mtune=ultrasparc</samp> option instead, which produces
   1233 code that, unlike full 64-bit code, can still run on non-UltraSPARC
   1234 machines.
   1235 </p>
   1236 <p>When configuring the GNU Multiple Precision Library (GMP), the MPFR
   1237 library or the MPC library on a Solaris 7 or later system, the canonical
   1238 target triplet must be specified as the <code>build</code> parameter on the
   1239 configure line.  This target triplet can be obtained by invoking <code>./config.guess</code> in the toplevel source directory of GCC (and
   1240 not that of GMP or MPFR or MPC).  For example on a Solaris 9 system:
   1241 </p>
   1242 <div class="smallexample">
   1243 <pre class="smallexample">% ./configure --build=sparc-sun-solaris2.9 --prefix=xxx
   1244 </pre></div>
   1245 
   1246 <hr />
   1247 <a name="sparc-sun-solaris210"></a><a name="sparc-sun-solaris2_002e10"></a>
   1248 <h3 class="heading">sparc-sun-solaris2.10</h3>
   1249 <p>There is a bug in older versions of the Sun assembler which breaks
   1250 thread-local storage (TLS).  A typical error message is
   1251 </p>
   1252 <div class="smallexample">
   1253 <pre class="smallexample">ld: fatal: relocation error: R_SPARC_TLS_LE_HIX22: file /var/tmp//ccamPA1v.o:
   1254   symbol &lt;unknown&gt;: bad symbol type SECT: symbol type must be TLS
   1255 </pre></div>
   1256 
   1257 <p>This bug is fixed in Sun patch 118683-03 or later.
   1258 </p>
   1259 <hr />
   1260 <a name="sparc-x-linux"></a><a name="sparc-*-linux*"></a>
   1261 <h3 class="heading">sparc-*-linux*</h3>
   1262 
   1263 <hr />
   1264 <a name="sparc64-x-solaris2"></a><a name="sparc64-*-solaris2*"></a>
   1265 <h3 class="heading">sparc64-*-solaris2*</h3>
   1266 <p>When configuring the GNU Multiple Precision Library (GMP), the MPFR
   1267 library or the MPC library, the canonical target triplet must be specified
   1268 as the <code>build</code> parameter on the configure line.  For example
   1269 on a Solaris 9 system:
   1270 </p>
   1271 <div class="smallexample">
   1272 <pre class="smallexample">% ./configure --build=sparc64-sun-solaris2.9 --prefix=xxx
   1273 </pre></div>
   1274 
   1275 <hr />
   1276 <a name="sparcv9-x-solaris2"></a><a name="sparcv9-*-solaris2*"></a>
   1277 <h3 class="heading">sparcv9-*-solaris2*</h3>
   1278 <p>This is a synonym for &lsquo;<samp>sparc64-*-solaris2*</samp>&rsquo;.
   1279 </p>
   1280 <hr />
   1281 <a name="c6x-x-x"></a><a name="c6x-*-*"></a>
   1282 <h3 class="heading">c6x-*-*</h3>
   1283 <p>The C6X family of processors. This port requires binutils-2.22 or newer.
   1284 </p>
   1285 <hr />
   1286 <a name="tilegx-*-linux"></a><a name="tilegx-*-linux*"></a>
   1287 <h3 class="heading">tilegx-*-linux*</h3>
   1288 <p>The TILE-Gx processor in little endian mode, running GNU/Linux.  This
   1289 port requires binutils-2.22 or newer.
   1290 </p>
   1291 <hr />
   1292 <a name="tilegxbe-*-linux"></a><a name="tilegxbe-*-linux*"></a>
   1293 <h3 class="heading">tilegxbe-*-linux*</h3>
   1294 <p>The TILE-Gx processor in big endian mode, running GNU/Linux.  This
   1295 port requires binutils-2.23 or newer.
   1296 </p>
   1297 <hr />
   1298 <a name="tilepro-*-linux"></a><a name="tilepro-*-linux*"></a>
   1299 <h3 class="heading">tilepro-*-linux*</h3>
   1300 <p>The TILEPro processor running GNU/Linux.  This port requires
   1301 binutils-2.22 or newer.
   1302 </p>
   1303 <hr />
   1304 <a name="visium-x-elf"></a><a name="visium-*-elf"></a>
   1305 <h3 class="heading">visium-*-elf</h3>
   1306 <p>CDS VISIUMcore processor.
   1307 This configuration is intended for embedded systems.
   1308 </p>
   1309 <hr />
   1310 <a name="x-x-vxworks"></a><a name="g_t*-*-vxworks*"></a>
   1311 <h3 class="heading">*-*-vxworks*</h3>
   1312 <p>Support for VxWorks is in flux.  At present GCC supports <em>only</em> the
   1313 very recent VxWorks 5.5 (aka Tornado 2.2) release, and only on PowerPC.
   1314 We welcome patches for other architectures supported by VxWorks 5.5.
   1315 Support for VxWorks AE would also be welcome; we believe this is merely
   1316 a matter of writing an appropriate &ldquo;configlette&rdquo; (see below).  We are
   1317 not interested in supporting older, a.out or COFF-based, versions of
   1318 VxWorks in GCC 3.
   1319 </p>
   1320 <p>VxWorks comes with an older version of GCC installed in
   1321 <samp><var>$WIND_BASE</var>/host</samp>; we recommend you do not overwrite it.
   1322 Choose an installation <var>prefix</var> entirely outside <var>$WIND_BASE</var>.
   1323 Before running <code>configure</code>, create the directories <samp><var>prefix</var></samp>
   1324 and <samp><var>prefix</var>/bin</samp>.  Link or copy the appropriate assembler,
   1325 linker, etc. into <samp><var>prefix</var>/bin</samp>, and set your <var>PATH</var> to
   1326 include that directory while running both <code>configure</code> and
   1327 <code>make</code>.
   1328 </p>
   1329 <p>You must give <code>configure</code> the
   1330 <samp>--with-headers=<var>$WIND_BASE</var>/target/h</samp> switch so that it can
   1331 find the VxWorks system headers.  Since VxWorks is a cross compilation
   1332 target only, you must also specify <samp>--target=<var>target</var></samp>.
   1333 <code>configure</code> will attempt to create the directory
   1334 <samp><var>prefix</var>/<var>target</var>/sys-include</samp> and copy files into it;
   1335 make sure the user running <code>configure</code> has sufficient privilege
   1336 to do so.
   1337 </p>
   1338 <p>GCC&rsquo;s exception handling runtime requires a special &ldquo;configlette&rdquo;
   1339 module, <samp>contrib/gthr_supp_vxw_5x.c</samp>.  Follow the instructions in
   1340 that file to add the module to your kernel build.  (Future versions of
   1341 VxWorks will incorporate this module.)
   1342 </p>
   1343 <hr />
   1344 <a name="x86-64-x-x"></a><a name="x86_005f64-*-*_002c-amd64-*-*"></a>
   1345 <h3 class="heading">x86_64-*-*, amd64-*-*</h3>
   1346 <p>GCC supports the x86-64 architecture implemented by the AMD64 processor
   1347 (amd64-*-* is an alias for x86_64-*-*) on GNU/Linux, FreeBSD and NetBSD.
   1348 On GNU/Linux the default is a bi-arch compiler which is able to generate
   1349 both 64-bit x86-64 and 32-bit x86 code (via the <samp>-m32</samp> switch).
   1350 </p>
   1351 <hr />
   1352 <a name="x86-64-x-solaris210"></a><a name="x86_005f64-*-solaris2_002e1_005b0-9_005d*"></a>
   1353 <h3 class="heading">x86_64-*-solaris2.1[0-9]*</h3>
   1354 <p>GCC also supports the x86-64 architecture implemented by the AMD64
   1355 processor (&lsquo;<samp>amd64-*-*</samp>&rsquo; is an alias for &lsquo;<samp>x86_64-*-*</samp>&rsquo;) on
   1356 Solaris 10 or later.  Unlike other systems, without special options a
   1357 bi-arch compiler is built which generates 32-bit code by default, but
   1358 can generate 64-bit x86-64 code with the <samp>-m64</samp> switch.  Since
   1359 GCC 4.7, there is also a configuration that defaults to 64-bit code, but
   1360 can generate 32-bit code with <samp>-m32</samp>.  To configure and build
   1361 this way, you have to provide all support libraries like <samp>libgmp</samp>
   1362 as 64-bit code, configure with <samp>--target=x86_64-pc-solaris2.1x</samp>
   1363 and &lsquo;<samp>CC=gcc -m64</samp>&rsquo;.
   1364 </p>
   1365 <hr />
   1366 <a name="xtensa-x-elf"></a><a name="xtensa*-*-elf"></a>
   1367 <h3 class="heading">xtensa*-*-elf</h3>
   1368 <p>This target is intended for embedded Xtensa systems using the
   1369 &lsquo;<samp>newlib</samp>&rsquo; C library.  It uses ELF but does not support shared
   1370 objects.  Designed-defined instructions specified via the
   1371 Tensilica Instruction Extension (TIE) language are only supported
   1372 through inline assembly.
   1373 </p>
   1374 <p>The Xtensa configuration information must be specified prior to
   1375 building GCC.  The <samp>include/xtensa-config.h</samp> header
   1376 file contains the configuration information.  If you created your
   1377 own Xtensa configuration with the Xtensa Processor Generator, the
   1378 downloaded files include a customized copy of this header file,
   1379 which you can use to replace the default header file.
   1380 </p>
   1381 <hr />
   1382 <a name="xtensa-x-linux"></a><a name="xtensa*-*-linux*"></a>
   1383 <h3 class="heading">xtensa*-*-linux*</h3>
   1384 <p>This target is for Xtensa systems running GNU/Linux.  It supports ELF
   1385 shared objects and the GNU C library (glibc).  It also generates
   1386 position-independent code (PIC) regardless of whether the
   1387 <samp>-fpic</samp> or <samp>-fPIC</samp> options are used.  In other
   1388 respects, this target is the same as the
   1389 <a href="#xtensa*-*-elf">&lsquo;<samp>xtensa*-*-elf</samp>&rsquo;</a> target.
   1390 </p>
   1391 <hr />
   1392 <a name="windows"></a><a name="Microsoft-Windows"></a>
   1393 <h3 class="heading">Microsoft Windows</h3>
   1394 
   1395 <a name="Intel-16-bit-versions"></a>
   1396 <h4 class="subheading">Intel 16-bit versions</h4>
   1397 <p>The 16-bit versions of Microsoft Windows, such as Windows 3.1, are not
   1398 supported.
   1399 </p>
   1400 <p>However, the 32-bit port has limited support for Microsoft
   1401 Windows 3.11 in the Win32s environment, as a target only.  See below.
   1402 </p>
   1403 <a name="Intel-32-bit-versions"></a>
   1404 <h4 class="subheading">Intel 32-bit versions</h4>
   1405 <p>The 32-bit versions of Windows, including Windows 95, Windows NT, Windows
   1406 XP, and Windows Vista, are supported by several different target
   1407 platforms.  These targets differ in which Windows subsystem they target
   1408 and which C libraries are used.
   1409 </p>
   1410 <ul>
   1411 <li> Cygwin <a href="#x-x-cygwin">*-*-cygwin</a>: Cygwin provides a user-space
   1412 Linux API emulation layer in the Win32 subsystem.
   1413 </li><li> MinGW <a href="#x-x-mingw32">*-*-mingw32</a>: MinGW is a native GCC port for
   1414 the Win32 subsystem that provides a subset of POSIX.
   1415 </li><li> MKS i386-pc-mks: NuTCracker from MKS.  See
   1416 <a href="https://www.mkssoftware.com">https://www.mkssoftware.com</a> for more information.
   1417 </li></ul>
   1418 
   1419 <a name="Intel-64-bit-versions"></a>
   1420 <h4 class="subheading">Intel 64-bit versions</h4>
   1421 <p>GCC contains support for x86-64 using the mingw-w64
   1422 runtime library, available from <a href="http://mingw-w64.org/doku.php">http://mingw-w64.org/doku.php</a>.
   1423 This library should be used with the target triple x86_64-pc-mingw32.
   1424 </p>
   1425 <p>Presently Windows for Itanium is not supported.
   1426 </p>
   1427 <a name="Windows-CE"></a>
   1428 <h4 class="subheading">Windows CE</h4>
   1429 <p>Windows CE is supported as a target only on Hitachi
   1430 SuperH (sh-wince-pe), and MIPS (mips-wince-pe).
   1431 </p>
   1432 <a name="Other-Windows-Platforms"></a>
   1433 <h4 class="subheading">Other Windows Platforms</h4>
   1434 <p>GCC no longer supports Windows NT on the Alpha or PowerPC.
   1435 </p>
   1436 <p>GCC no longer supports the Windows POSIX subsystem.  However, it does
   1437 support the Interix subsystem.  See above.
   1438 </p>
   1439 <p>Old target names including *-*-winnt and *-*-windowsnt are no longer used.
   1440 </p>
   1441 <p>PW32 (i386-pc-pw32) support was never completed, and the project seems to
   1442 be inactive.  See <a href="http://pw32.sourceforge.net/">http://pw32.sourceforge.net/</a> for more information.
   1443 </p>
   1444 <p>UWIN support has been removed due to a lack of maintenance.
   1445 </p>
   1446 <hr />
   1447 <a name="x-x-cygwin"></a><a name="g_t*-*-cygwin"></a>
   1448 <h3 class="heading">*-*-cygwin</h3>
   1449 <p>Ports of GCC are included with the
   1450 <a href="http://www.cygwin.com/">Cygwin environment</a>.
   1451 </p>
   1452 <p>GCC will build under Cygwin without modification; it does not build
   1453 with Microsoft&rsquo;s C++ compiler and there are no plans to make it do so.
   1454 </p>
   1455 <p>The Cygwin native compiler can be configured to target any 32-bit x86
   1456 cpu architecture desired; the default is i686-pc-cygwin.  It should be
   1457 used with as up-to-date a version of binutils as possible; use either
   1458 the latest official GNU binutils release in the Cygwin distribution,
   1459 or version 2.20 or above if building your own.
   1460 </p>
   1461 <hr />
   1462 <a name="x-x-mingw32"></a><a name="g_t*-*-mingw32"></a>
   1463 <h3 class="heading">*-*-mingw32</h3>
   1464 <p>GCC will build with and support only MinGW runtime 3.12 and later.
   1465 Earlier versions of headers are incompatible with the new default semantics
   1466 of <code>extern inline</code> in <code>-std=c99</code> and <code>-std=gnu99</code> modes.
   1467 </p>
   1468 <hr />
   1469 <a name="older"></a><a name="Older-systems"></a>
   1470 <h3 class="heading">Older systems</h3>
   1471 <p>GCC contains support files for many older (1980s and early
   1472 1990s) Unix variants.  For the most part, support for these systems
   1473 has not been deliberately removed, but it has not been maintained for
   1474 several years and may suffer from bitrot.
   1475 </p>
   1476 <p>Starting with GCC 3.1, each release has a list of &ldquo;obsoleted&rdquo; systems.
   1477 Support for these systems is still present in that release, but
   1478 <code>configure</code> will fail unless the <samp>--enable-obsolete</samp>
   1479 option is given.  Unless a maintainer steps forward, support for these
   1480 systems will be removed from the next release of GCC.
   1481 </p>
   1482 <p>Support for old systems as hosts for GCC can cause problems if the
   1483 workarounds for compiler, library and operating system bugs affect the
   1484 cleanliness or maintainability of the rest of GCC.  In some cases, to
   1485 bring GCC up on such a system, if still possible with current GCC, may
   1486 require first installing an old version of GCC which did work on that
   1487 system, and using it to compile a more recent GCC, to avoid bugs in the
   1488 vendor compiler.  Old releases of GCC 1 and GCC 2 are available in the
   1489 <samp>old-releases</samp> directory on the <a href="../mirrors.html">GCC mirror
   1490 sites</a>.  Header bugs may generally be avoided using
   1491 <code>fixincludes</code>, but bugs or deficiencies in libraries and the
   1492 operating system may still cause problems.
   1493 </p>
   1494 <p>Support for older systems as targets for cross-compilation is less
   1495 problematic than support for them as hosts for GCC; if an enthusiast
   1496 wishes to make such a target work again (including resurrecting any of
   1497 the targets that never worked with GCC 2, starting from the last
   1498 version before they were removed), patches
   1499 <a href="../contribute.html">following the usual requirements</a> would be
   1500 likely to be accepted, since they should not affect the support for more
   1501 modern targets.
   1502 </p>
   1503 <p>For some systems, old versions of GNU binutils may also be useful,
   1504 and are available from <samp>pub/binutils/old-releases</samp> on
   1505 <a href="https://sourceware.org/mirrors.html">sourceware.org mirror sites</a>.
   1506 </p>
   1507 <p>Some of the information on specific systems above relates to
   1508 such older systems, but much of the information
   1509 about GCC on such systems (which may no longer be applicable to
   1510 current GCC) is to be found in the GCC texinfo manual.
   1511 </p>
   1512 <hr />
   1513 <a name="elf"></a><a name="all-ELF-targets-_0028SVR4_002c-Solaris-2_002c-etc_002e_0029"></a>
   1514 <h3 class="heading">all ELF targets (SVR4, Solaris 2, etc.)</h3>
   1515 <p>C++ support is significantly better on ELF targets if you use the
   1516 <a href="./configure.html#with-gnu-ld">GNU linker</a>; duplicate copies of
   1517 inlines, vtables and template instantiations will be discarded
   1518 automatically.
   1519 </p>
   1520 
   1521 <hr />
   1522 <p>
   1523 <p><a href="./index.html">Return to the GCC Installation page</a>
   1524 </p>
   1525 
   1526 
   1527 
   1528 
   1529 
   1530 
   1531 </body>
   1532 </html>
   1533